Finding the Average Color within a Polygon Bound in OpenCV

Question

Background

I am trying to create a utility that will return the average pixel color within a given polygon using OpenCV. The polygon will be defined via 4 points, but it is not necessarily a rectangle/square. For example, the following structures are to be expected:

     A__________B    A_______B
    /          /     \       \
   /          /       \       \
 D/__________/C       D\_______\C

---

Given a cv::Mat image in OpenCV and a polygon defined by the points (A, B, C, D). I know points A, B, C, and D, but I want to calc the average pixel color within the polygon. I wanted to get some suggestions from the OpenCV community on how to do this most efficiently.

---

Research Done

Another post on StackOverflow suggested drawing contours using the drawContours function and then take the mean of the bounding rectangle surrounding the contour. I would obviously have to modify the mean calculation so that it uses the polygons drawn by the fillPoly function instead.

Suggestions/Concerns are much appreciated!

Answer 1

You can simply use the mean function with a mask, where the mask is your filled polygon.

#include <opencv2\opencv.hpp>
#include <iostream>
using namespace cv;

int main()
{
    // Create a black image with a gray rectangle on top left
    Mat1b img(300, 300, uchar(0));
    rectangle(img, Rect(0, 0, 100, 100), Scalar(100), CV_FILLED);

    // Define a polygon
    Point pts[1][4];
    pts[0][0] = Point(20, 20);
    pts[0][1] = Point(40, 100);
    pts[0][2] = Point(200, 60);
    pts[0][3] = Point(150, 30);

    const Point* points[1] = {pts[0]};
    int npoints = 4;

    // Create the mask with the polygon
    Mat1b mask(img.rows, img.cols, uchar(0));
    fillPoly(mask, points, &npoints, 1, Scalar(255));

    // Compute the mean with the computed mask
    Scalar average = mean(img, mask);

    std::cout << average << std::endl;

    return 0;
}